Refusing the insurable, accepting what seems uninsurable
Tomorrow’s risk is never fully written in yesterday’s data.
Sometimes I turn down risks that are clearly insurable. And conversely, I accept risks that are barely documented.
Over four years working on new product underwriting at Seyna, I have spoken with more than 100 brokers across more than 250 projects, from travel to health, breakdown/damage/theft, and pet health insurance. Despite the variety of projects, this paradox keeps coming up.
What do I mean here by an insurable risk?
My actuary instinct tells me it is a risk for which I can use historical data to estimate a reliable pure premium. Yet even with data available, we can never be certain it reflects tomorrow’s risk: will the distribution channel be the same as the one that generated the historical record? What about the claims manager? The target customer base? The sales trajectory?
On these points, data tells us very little. It is the conversations with the broker that reveal whether tomorrow’s risk resembles yesterday’s. And that is where the paradox plays out: a file with a strong track record can conceal real ambiguity, while a file that is almost entirely data-free can be very well structured.
Why I sometimes decline an insurable risk
What I am looking for when co-creating a white-label product with a broker is a launch that makes sense for both parties and creates real value. Even when a risk looks insurable on paper, there are things to understand before committing.
The first: who will distribute the product, and how? When the answer is “several parties are interested,” with no name attached, I start to wonder whether the product addresses a precise market need at all. We risk endless back-and-forth on the design, wasted time on both sides, and costs, often underestimated, for a product that may launch very late, if ever. And without an identified distributor, it is hard to answer the questions raised above: which channel, which customer base, what volumes? I have no basis on which to estimate the risk.
The second: what is the business plan and the realistic potential of the product? A business plan is a business plan, with all the uncertainty that entails. But when the projected conversion rate is well above anything observed in the market, it is worth digging in and asking questions. This matters because if volumes diverge significantly from projections, the product’s financial balance will too.
The third is more surprising: a file that looks too perfect is not necessarily the most reassuring. A pack of aggregated data showing very low historical claims and no flagged concerns, for instance, actually makes me want to look harder. Conversely, when the broker tells me upfront what worries them about their own file, that is reassuring, and we can work together to find a lasting solution. It is also an opportunity to test our capacity for co-construction and to build a long-term partnership capable of weathering difficult moments.
Reading these signals is obviously not an exact science. I get it wrong, and often the Seyna sales team would be quick to point that out 😉, and that is part of what makes underwriting interesting: making decisions with the information you have, without ever having all the answers.
Why I sometimes accept a risk that seems uninsurable
Conversely, I have said yes to risks I was initially sceptical about: a guarantee where the potential frequency could legitimately worry an insurer, and no historical data to provide reassurance.
What changed everything was the way the broker presented it. They explained precisely how they planned to monitor their claims experience and respond if it drifted, adjusting distribution, pricing, or the product itself. They could answer the questions raised earlier: channel, customer base, sales trajectory. They were also transparent about what they did not yet know, and about how they intended to find out quickly.
What I had initially thought barely insurable became a risk whose uncertainty could be tracked and managed: on our side, we monitor our portfolios closely, and on the other side there was someone ready to react alongside us if problems emerged.
Ultimately, an insurable risk is not simply a risk we know how to estimate. It is a risk we know how to manage together.
